🚀 Agentic-Notify
Agentic-Notify is an open-source Python SDK for building agentic, notification-driven workflows. It acts as an autonomous orchestration layer that intercepts mobile or web notifications, plans multi-step actions using Large Language Models (LLMs), and executes them deterministically using strict Pydantic schemas.
📖 Full Documentation
Visit the official documentation site at: https://nikhilkanamadi.github.io/agentic-notify/
⚡ The Problem It Solves
Traditional notification systems (like Firebase Cloud Messaging or APNs) are simple delivery pipes. They push a string to your phone, and you must manually open the app to take action.
Agentic-Notify transforms notifications into autonomous workflows. Instead of just displaying "Your flight is delayed," the engine intercepts the notification, uses a Dynamic Policy Engine to check your preferences, queries an LLM Planner, and executes a multi-step workflow (e.g., canceling your Uber, emailing your boss, and setting a native reminder).
📦 Installation
To install the library and the optional backend engine (FastAPI):
pip install agentic-notify[fastapi]
🛠️ Quick Start Guide
The easiest way to understand the architecture is to see it run end-to-end. The library is completely modular. You initialize the Orchestrator, register your Adapters (tools), and define the Workflow JSON graph.
1. Initialize the Engine
import asyncio
from agentic_notify.orchestrator import NotificationOrchestrator
from agentic_notify.storage.memory import InMemoryStore
from agentic_notify.policies.engine import PolicyEngine
storage = InMemoryStore()
policy_engine = PolicyEngine()
orchestrator = NotificationOrchestrator(storage=storage, policy_engine=policy_engine)
2. Register Your Adapters (Tools)
Adapters are standard python classes that execute real-world logic.
from agentic_notify.adapters.notes import NotesAdapter
from agentic_notify.adapters.mock_reminder import MockReminderAdapter
orchestrator.register_adapter(NotesAdapter())
orchestrator.register_adapter(MockReminderAdapter())
3. Define the Agentic Workflow
Using strict Pydantic models, define the graph that the LLM or Routing Engine will follow.
from agentic_notify.schemas.workflow import WorkflowDefinition, WorkflowTrigger, WorkflowStep
workflow = WorkflowDefinition(
workflow_id="wf_e2e_demo",
workflow_name="E2E Notification Handler",
trigger=WorkflowTrigger(type="notification"),
steps=[
WorkflowStep(
id="step_save_note",
kind="adapter",
name="save_note",
input_from={"title": "event.title", "content": "event.body"}
),
WorkflowStep(
id="step_create_reminder",
kind="adapter",
name="create_reminder",
input_from={"task_title": "event.title", "task_text": "event.body"}
)
]
)
orchestrator.register_workflow(workflow)
# Route incoming events with a specific keyword to this workflow
orchestrator.router.add_rule(lambda event: "charge" in str(event.title).lower(), "wf_e2e_demo")
4. Inject a Mock Notification Event
The orchestrator converts raw device payloads into unified NotificationEvent schemas.
mock_payload = {
"event_id": "evt_demo_001",
"source_platform": "ios",
"source_app": "finance",
"title": "Unusual Charge Detected",
"body": "A charge of $500.00 was detected. Please review.",
"received_at": "2026-03-28T12:00:00Z"
}
async def run():
result = await orchestrator.process_event(mock_payload)
print(result)
asyncio.run(run())
🧠 Advanced Features Included
- Pydantic Data Contracts: Zero LLM hallucinations. The engine forces Agentic output into strict JSON schema validation locks.
- Human-in-the-Loop Hooks: The
ApprovalHandlersafely halts execution on high-risk actions (like payments) and suspends state until human authorization is granted. - Durable Postgres Storage: Fully tracks all state execution securely by
tenant_idfor resumable, safe workflows. - Dynamic Policy Engine: A middleware barrier that checks dynamic user-profiles and boundaries before any action is ever triggered.
Metadata
Release files for agentic-notify 0.1.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| agentic_notify-0.1.1.tar.gz | 15.8 kB | Without provenance | Details for agentic_notify-0.1.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| agentic_notify-0.1.1-py3-none-any.whl | Python 3 | none | any | Details for agentic_notify-0.1.1-py3-none-any.whl |
Total release size: 42.7 kB
Release files / agentic_notify-0.1.1.tar.gz
| Download URL | agentic_notify-0.1.1.tar.gz |
|---|---|
| Size | 15.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
2662b312794d4de3def4b834c28db4eeedb5c24c7ba9f63ab5d59851e688ff91
|
|
BLAKE2b-256 checksum How to use checksums |
5d7bfeae3f455adeb6d5f5bcb8a82117d147356d277bb4f843573f95526d24ce
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.4
|
Release files / agentic_notify-0.1.1-py3-none-any.whl
| Download URL | agentic_notify-0.1.1-py3-none-any.whl |
|---|---|
| Size | 26.9 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
8469feec7d66cb1077d9b3b2e7b41762fc6202e3245ad2a21d044ea736316652
|
|
BLAKE2b-256 checksum How to use checksums |
40e6cfc04bb486480de1cc808094927cf70e0c338a2f51ebaa4299d28d874ccd
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.4
|
Security for agentic-notify 0.1.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.